Skip to main content

Full text of "SCHS statistical primer"

See other formats


state Center for Health and Environmental Statistics 

P.O. Box 29538 • Raleigh, NC 27626-0538 




No. 11 

October 1993 



Kathryn B. Surles 

Recent studies'-^ examine the health status of Black and Native American residents of North 
Carolina. A similar study is desired for Hispanics, one of the fastest-growing segments of the state ' s 
population. However, data obstacles prevent such analysis at this time. 

The intent of this report is to describe the data problems identified and to suggest some possible 
remedies. It is hoped that this background will prove useful to those concerned with the health and 
health care of Hispanics. 


Some data sets pertaining to the status of individuals do not include information on Hispanic origin. 
In North Carolina, for example, it is not included on the induced abortion report form, which is used 
to compute abortion measures as well as pregnancy rates. Although the communicable disease report 
card includes Hispanic origin (yes/no) since 1987, the variable has not been included on data files 
routinely acquired by the State Center. 

Among data files of the N.C. Division of Social Services, Hispanic origin (yes/no) is included on 
these: food stamps, financial eligibility (Medicaid, AFDC, etc.), nonfinancial services (foster care, day 
care, etc.), and aging. Neither race nor Hispanic origin is collected on claims files. 

Race and Hispanic origin also are not included on hospital discharge records collected by the N.C. 
Medical Database Commission. Hispanic origin (yes/no) is collected by the state's Behavioral Risk 
Factor Surveillance System and Youth Risk Behavior Survey but not by other health-related surveys 
conducted in recent years. In all survey efforts, Hispanics must be oversampled to obtain stable 



The U.S. Census collects information on Hispanic origin and Hispanic subgroups as do North 
Carolina birth and death certificates since 1988. However, some of these data have substantial 
problems as described below. 

1990 Census Counts 

North Carolinians age five and older who reported Hispanic origin in the 1990 census numbered 
67,936, while the number of persons age five and older who reported speaking Spanish or Spanish 
Creole at home numbered 105,963. This discrepancy suggests an undercount of Hispanics. Thus, 
population-based rates of birth, death, disease, etc. would be expected to be biased upward for North 
Carolina's Hispanic population. The undercount of Hispanics appears to be especially great at younger 
ages as shown by available age-specific counts and those ratios: 

Age in 1990 

Pgrsftns Rgporting 5 and Older s-n 18-64 65 and Older 

Hispanic Origin* 67,836 15,067 49,963 2,806 

Spanish Speaking* 105,963 31,276 71,252 3,435 

Ratio .64 .48 .70 .82 

♦Hispanic Origin is ftom Tape File 1 (100% tabulations) while Spanish Speaking is from Tape 
File 3 (sample tabulations). 

In addition to the above classification problem, special problems exist in data collection among 
migrant workers who move often and undocumented aliens who fear discovery. The extent to which 
Hispanic persons were missed entirely in the 1990 census is unknown. 

Live Births 

Usually provided by the newbom's mother, race/ethnicity information at birth should be relatively 
accurate. Thus, given the census problems discussed above, it is not surprising that Hispanics represent 
a substantially higher proportion of live births than they do census counts of females of childbearing 
ages, as shown below. The result is that 1990 Hispanic birth rates would probably be biased upward 
to an unknown extent 

Age in 1990 
Hispanics as a Total 

Percentage of 1990 If^ X^iM 2S=M .2fiM 

Live Births 1.68 1.90 1.51 1.34 

Female Population 1.11 1.27 1.22 0.82 

Ratio 1.51 1.50 1.24 1.63 

Age in 1990 


Under 20 20-64 

65 and Older 


1.03 0.43 



1.50 1.16 



Race/ethnicity of a decedent sometimes is only the judgement call of a funeral director, and Hispanics 
appear to be undercounted on death certificates. The percentages and ratios below suggest that 1990 
death rates for Hispanics would be biased downward despite census undercounts (denominators) 
which would serve to drive the rates up. 

Hispanics as a 
Percentage ofl 990 


Ratio 0.18 0.69 0.37 0.23 

Medical Examiner Deaths 

North Carolina law requires that all deaths suspected to be due to a violent or traumatic injury or 
accident be investigated by a Medical Examiner. Certain other categories of deaths also fall under the 
purview of the Medical Examiner — deaths that are medically unattended, that occur during a surgical 
procedure, that arc due to suspicious circumstances, or where the death is sudden or not related to 
known previous disease. 

For these deaths, Hispanic origin (yes/no) was added to the detailed "Report of Investigation by 
Medical Examiner" in 1990. Thirty deaths (0.35%) were reported as Hispanic in 1990; 98 deaths 
(1.17%) were reported as Hispanic in 1 99 1 . The increase may reflect improved reporting performance 
after the initial year. 

Matched Birth/Infant Deaths 

Based on historical data, funeral directors appear more apt now than formerly to ascribe to a 
deceased infant the same race as reported by the mother at the infant' s birth. However, for birth years 
1988-91, funeral directors reported as Hispanic (usually Mexican) 18 infant decedents who were 
considered non-Hispanic at birth. Seven of these could be cases of a non-Hispanic mother and an 
Hispanic father, in 1990, North Carolina followed the National Center for Health Statistics in ascribing 
to a newborn the race/ethnicity of the mother regardless of that of father. 

At the same time, funeral directors reported as non-Hispanic 13 infant decedents who were 
considered Hispanic (usually Mexican) at birth. The 18 above and 13 here compare to only 24 infant 
decedents who were considered Hispanic at both birth and death. 

From these data, it is obvious that discordances between ethnicity at birth versus death will affect 
the infant death data for Hispanics. For analyses by race, it is customary to use race at death in 
computations by death year but race at birth in computations by year of birth. The latter involve use 
of the matched birth/infant death file in order to examine birth characteristics associated with infant 


Census undercounts notwithstanding, those data arc used below to get a general idea of the mix of 
Hispanics in North Carolina. Results suggest, as expected, a mixture of ethnic, cultural, and racial 
origins with no particular group dominating the picture. Thus, aggregate Hispanic health data may 
not be all that helpful to persons planning health care delivery at the local level. A recent study in 
Washington State^ reveals contrasting health risks among Hispanic subgroups. 







Puerto Rican 
Other Hispanic* 












American Indian, Eskimo, Aleut 

Asian or Pacific Islander 

Other Race** 





•Persons reporting "Other Spanish/Hispanic" origin from Spain, the Spanish-speaking 
countries of Central or South America or the Dominican RepubUc, or persons of Hispanic 
origin identifying themselves generally as Spanish, Spanish- American, Hispanic, Hispano, 
Latino, etc. 

**PCTSons reporting in the "Other race" category and providing write-in entries such as multi- 
racial, multi-ethnic, mixed, interracial, Wesort, or a Spanish/Hispanic origin group (such 
as Mexican, Cuban, or Puerto Rican). 


Census Data 

Accurate census counts for population-based rates are crucial to the analysis of health data. 
Unfortunately, an undercount of Hispanics appears to have occurred in North Carolina in 1990, and 
there appears no satisfactory remedy for this situation. Certainly, it is hoped that the Census Bureau 
will take whatever steps are necessary to improve the situation in the 2000 census. 

For programs concerned with the geographic locations and other demographics of people who 
primarily speak Spanish, help may be available through use of summary tape files maintained by the 
census. From these tapes, one may obtain counts of persons who reported speaking Spanish at home. 
Counts are available by age, race, and county. 

For other breakdowns of persons who reported speaking Spanish at home, one may use machine- 
readable data files maintained by the census. These Ihiblic Use Microdata Samples (PUMS) contain 
records representing 5% and 1% samples of the housing units in the U.S. and are available for each 
state. The file contains "housing" records, each followed by a variable number of "person" records, 
one for each occupant of the housing unit. Although these files are extremely large, they can be handled 
by most statistical or report-writing software."* 

The PUMS files contain numerous data items including each person's ability to speak English, 
language spoken at home, age, and relationship to the designated householder. These data items may 
allow the user to construct a sample of households and/or persons who speak Spanish at home and 
do not speak English well. The item, "telephone in housing unit," may also be useful in conducting 
telephone surveys that oversample those who predominantly speak Spanish. 

Note: As a first step in using the PUMS files, one should verify estimates 
fi-om PUMS by applying the individual weights provided on PUMS 
and comparing tabulated results to published estimates fi-om the 
census' summary tape file 3, available from the State Data Center, 
Office of State Planning, at (919) 733-4131. 

State Reporting Systems 

The State Center will add an item on Hispanic origin and Hispanic subgroups to its induced abortion 
report form, effective January 1, 1994. The State Center will also add Hispanic origin to its 
computerized communicable disease files beginning with data year 1 987 and encourages the addition 
of Hispanic subgroups to the communicable disease report card. Hispanic subgroups should also be 
added to the Central Cancer Registry Incidence Report and the Report of Investigation by Medical 
Examiner, and Hispanic origin and subgroups should be added to the lead poisoning database of the 
State Laboratory. The Medical Database Commission is likewise urged to seek legislation to allow 
including race and Hispanic origin/subgroups in its hospital discharge reporting system as well as 
future nonhospital reporting systems as they develop. 

Food stamps, Medicaid, and other programs of the Division of Social Services do include Hispanic 
origin but not Hispanic subgroups. These and other state programs need to be told of the need for 
these data although lack of "administrative necessity" may undermine a recommendation to add this 
tevel of detail. 

Although the state's birth, death, and fetal death files include Hispanic origin and subgroups since 
1988, it appears that Hispanic deaths are underreported, especially at older ages. The State Center, 
working with the Vital Records Section, needs to do its part in raising the awareness level of funeral 
directors and other data providers concerning the need for this information. 

The examination of data from the North Carolina birth certificate suggests that the Hispanic origin 
information may be used to examine the incidence of sociodemographic and medical risk factors 
among Hispanic mothers and their infants. These data may be examined separately for Mexicans, 
Puerto Ricans, Central/South Americans, and other Hispanics although the statistical problem of small 
numbers may prevent some analyses. These data for 1988-92 are being used to prepare a fact sheet 
on the maternal and infant health indicators of the state's Hispanic residents. 

The North Carolina Birth Defects Registry combines data from several separate sources to provide 
estimates of birth defects incidence in the state. Similarly, the Child Find File combines data from 
several early intervention programs to provide estimates of the number of children with or at risk for 
developmental delay. Both the defects registry and the child find file include information from the birth 
certificate, so estimates for Hispanics and Hispanic subgroups will be available following additional 
years of data collection. At least two more years arc needed to build the numbers in order to provide 
reliable estimates for Hispanics. 

Meanwhile, other potential sources of Hispanic data are the Pregnancy Nutrition Surveillance 
System, the Supplemental Food Program for Women, Infants, and Children (WIQ, and the Health 
Services Information System (HSIS). WIC has recently merged with HSIS, and there are plans to 
routinely match HSIS prenatal records to the corresponding birth certificates. HSIS is an automated 
reporting system used by local health departments and special contract providers (over 700 sites) to 
report on the delivery of public health services. 

Note: In using data from the birth certificate and fetal death report, the user 
should tabulate on Hispanic origin of the mother, in accordance with 
national procedures for race since 1989. Origin of the newborn' s father 
would not be considered. 

Sample Surveys 

North Carolina participates with the Centers for Disease Control (CDC) in an annual telephone 
interview survey (Behavioral Risk Factor Surveillance System) that solicits information about the 
health behaviors of adults 18 and older. Post-survey adjustments serve to make the results 
representative of all adults, not just those having telephones. Beginning in March 19 93jth£_North 
jCM5lina.Department«f^!ubliGJiistQictionJi^^ in a biennial Youth 

Risk Behavior Survey of 9th- 12th grade students. It is recommended that^othoflhesenafioiial 

surveys add Hispanic origin and Hispanic subgroups to their data collection instruments and 
oversample Hispanics and subgroups in order to obtain stable estimates for this vulnerable population. 
Likewise, all health-related surveys should collect Hispanic origin and subgroups and oversample 

On a very basic level, as suggested in the Washington State report,' the collectors of health survey 
data should examine whether the questions asked and the measures used to assess health status are 
appropriate for a particular minority group — in this case, Hispanics and those subgroups. 


This report points up a number of problems that must be solved before health data for Hispanics are 
generally available. Meanwhile, a fact sheet for Hispanics and Hispanic subgroups examines and 
compares data from the North Carolina birth certificate for birth years 1988-92. From this work, it 
is seen that health risks do vary among the Hispanic subgroups so that aggregate data for Hispanics 
as a whole will not suffice. All data collection instruments should collect Hispanic subgroup, and all 
programs should strive to raise the awareness level of providers conceming the need for this 

Meanwhile, it is hoped that steps will be taken to ensure the respondent's understanding of the race 
and Hispanic origin questions on the Year 2000 census form. Misunderstanding may have contributed 
to some misreporting of Hispanics in 1990. 

A recent report by the Centers for Disease Control and Prevention (CDC)' summarizes a workshop 
that addressed the role of race and ethnicity in public health surveillance. This work and its 
recommendations should be reviewed by those concerned with the important matters of minority 


1. N.C. Department of Environment, Health, and Natural Resources. "Health Status of Native 
Americans in North Carolina," CHES Studies, No. 75. Raleigh, August 1993. 

2. N.C. Department of Environment, Health, and Natural Resources. "Health Status of Blacks in 
North Carolina," CHES Studies. No. 76. Raleigh, October 1993. 

3. Washington State Department of Health, Center for Health Statistics. Washington State Health 
Data Report on People of Color. Olympia, October 1992. 

4. Census ofPopulation and Housing, 1990: Public Use Microdata Samples (machine-readable data 
files). Prepared by the Bureau of the Census. Washington: The Bureau (producer and 
distributor), 1992. 

5. Centers for Disease Control and Prevention. "Use of Race and Ethnicity in Public Health 
Surveillance: Summary of the CDC/ATSDR Workshop," Morbidity and Mortality Weekly 
Report . Vol. 42, No. RR- 10. U.S. Government Printing Office, June 25, 1993. 


I II nil 111 
3 3091 00747 1535 

Department of Environment, Health, and Natural Resources 
State Center for Health and Environmental Statistics 
P.O. Box 29538 
Raleigh, N.C. 27626-0538 


U.S. Postage 


Raleigh, N.C. 27626-0538 

Permit No. 1862 






1,200 copies of this public document were printed at a cost of $100.56 or 8.40 per copy.